You are on page 1of 51

# 1

Chapter 1

Introduction

When non economists want to make fun of economists (or when economists
want to make fun of each other), they often tell the following story:
A shipwreck has left a physicist, a chemist, and an economist without food on a
deserted island. A few days later a can of beans washes up on the shore. The physicist
proposes the following method of opening the can:
I've calculated the terminal velocity of a one-pound object--the weight of the
can-- thrown to a height of twenty feet is 183 feet per second. If we place a rock
under the can the impact should just burst the seams without spilling the beans.
The chemist's response is:
That's risky since we can't be sure we will throw it to the correct height. I've got a better
idea. Let's start a fire and heat the can on the coal for one minute, thirty-seven seconds. I've
calculated that this should just burst the seams. This method is less risky since we can
always push the can off the fire if it starts to burst sooner.

The economist's reaction is:
Both of your methods may work, but they are too complicated. My approach is
much simpler: Assume a can opener.
If you have studied economics before, you will appreciate the relevance of this
joke (and probably already have heard it more than once). If you have not studied
economics, you will soon learn why you should laugh harder than you did. The can
opener story illustrates one important truth and one important lie about economists.
The truth is that they approach problems by making assumptions (though,
unfortunately, this is not always a lie).

2

The Role of Assumptions
Economists make assumptions for the obvious reason that the world, viewed
economically, is too complicated to understand without some abstraction. To see this
point in a legal context, consider the well-known products liability case of Escola v.
Coca Cola Bottling Co., in which the plaintiff was injured by an exploding bottle of
soda. In a concurring opinion, Justice Traynor of the California Supreme Court made
the following remarks:1

3

I concur in the judgment [for the plaintiff], but i believe the manufacturer's
negligence should no longer be singled out as the basis of the plaintiff's right to
recover in cases like the present one. In my opinion it should now be recognized that
a manufacturer incurs an
absolute liability when an article that he has placed on
the market, knowing that it is to be used without inspection, proves to have a defect
that causes injury to human beings....Even if there is no negligence, ... public policy
demands that responsibility be fixed wherever it will most effectively reduce the
hazards to life and health inherent in defective products that reach the market. It is
evident that the manufacturer can anticipate some hazards and guard against
the
recurrence of others, as the public cannot. Those who suffer injury from
defective
products are unprepared to meet its consequences. The cost of injury and the loss of
time or health may be an overwhelming misfortune to the person injured,
1. Escola v. Coca Cola Bottling Co., 24 Cal 2d 453, 461-462, 150 P.2 436,
440-441(1944).
and a needless one, for the risk of injury can be insured by the manufacturer
and distributed among the public as a cost of doing business. It is to the public
interest to discourage the marketing of products having defects that are a menace to
the public. If such products nevertheless find their way into the market it is to the
public interest to place the responsibility for whatever injury they may cause upon the
manufacturer, who, even if he is not negligent in the manufacture of the product, is
responsible for its reaching the market. However intermittently such injuries may
occur and however haphazardly they may strike, the risk of their occurrence is a
constant risk and the general one. Against such a risk there should be general and
constant protection and the manufacturer is best situated to afford such protection.

4

This excerpt from Justice Traynor's opinion implicitly raises several basic
questions but does not go very far toward answering them. How is the criterion of
"public policy" or "the public interest" to be defined? Should it take into account
consideration of both efficiency and equity? What are the effect of the rules of
negligence and absolute (or strict) liability on the care exercised by manufacturers in
designing and producing products? On the care exercised by consumers in using
products? On the price and output of the industry? Do the answers to these questions
depended on whether the consumer miss-perceives the product risks? On whether the
victim is a third party rather than a consumer of the product? Should the
manufacturer be allowed to raise as a defense that the victim was contributorily
negligent in the use of the product? Who is the better bearer of the product risks that
are not eliminated—the manufacturer or the consumer? How does the answer to this
question depend on whether the manufacturer can self-insure or purchase liability
insurance? On whether the consumer can purchase first-party accident insurance or is
covered by general social insurance programs?
Unless you are already familiar with the economics literature on products on
liability, you probably cannot answer many of these questions in a systematic way.
By the end of this book however, you should be able to answer all of them by
thinking about them like and economist.
The way an economist would go about answering these questions would be to
isolate one or two of them at a time by making simplifying assumptions that
eliminate the others. For example, one might start with the case in which the victim is
a consumer of the product and assume that she has perfect information about the
product risks, that she does not affect the probability or magnitude of the harm by her
own actions, and that she is "risk neutral."*2 It is relatively straightforward in this
framework to determine what effects the rules of negligence and strict liability will
have on the care exercised by manufacturers and on the price and output of the
industry. One then could investigate the consequences of adding the following
complications to this framework: consumer misperceptions, joint determination by
the manufacturer and the consumer of the probability and magnitude of harm, and
risk aversion of the consumer (and possibly of the manufacturer). 3 The last
complication, risk aversion, would have to be considered both when insurance was
and was not assumed to be available.
Each of the special cases just described—corresponding to the a particular set
of assumptions—can be analyzed relatively easily because only a few issues are
considered at a time. Even though each special case is admittedly unrealistic by itself,
it will generate some insights that are relevant to real product liability problems. And
by examining all of the cases, one can obtain a comprehensive—and, more
importantly, comprehensible—perspective on the economics of the product liability
rules.
This discussion of the role of simplifying assumptions is the economic analysis
of product liability rules applies to every problem addressed by economists—

5

including every topic considered in this book.
The art of economics is picking assumptions that simplify a problem enough to better
understand certain features of it, without causing those features to be unimportant
ones. That economists are sometimes (often?)
2.The term”risk neutral” is defined at p.31 below
3.The concept of risk aversion is discussed at p. 57 below.
Thought to eliminate what is interesting about a problem through their assumptions is
what gives the can opener story its bite. ###
What has been said thus far is meant to be a warning, not an apology. Be
prepared to accept(at least for while) some obvious unrealistic assumptions. By the
end of this book, I hope to convince you that there are many insights to be gained
from the economic analysis of law and that these insights are the result of the artful
choice of simplifying assumptions.
The plan of this book is somewhat unconventional. Chapters introducing basic
economic concepts are interwoven with chapters applying those concepts to legal
problems. In this manner , the relevant economic ideas are developed systematically
for the readers not trained in economics, and readers with some training in economics
easily can identify and skip the materials with which they are familiar. The economic
concepts that are discussed include efficiency and equity, risk bearing and insurance,
and competitive markets. These concepts are applies to nuisance law. Breach of
contract, automobile accidents, law enforcement, pollution control, product liability,
principal-agent liability, litigation, and regulation. Also, in the concluding chapter,
some problems with the practical implementation of economic approach to law are
considered.
The subject matter is introduce in stages of increasing complexity, both to
simplify the exposition and to better convey the style of economic analysis. For
example, breach of contract remedies are first examined in Chapter 5 in a contractual
relationship I which the parties are assumed to be risk neutral. This chapter analyzes
the effects of the remedies on the parties' breach decisions on their”reliance”
expenditures. Then, in Chapter 7, the concept of risk aversion is introduced (and the
function of insurance is discussed ) in Chapter 8, breach of contract are reexamined
under the more realistic assumptions that the parties may be risk averse. That chapter
focuses on the effects and remedies on the allocation of risk of breaches that occur.
By developing the analysis in these stages, it is easier to see the separate economic
functions of breach of contract remedies to control the behavior of contracting
parties with respect to breach and reliance decisions, and to allocate the risk of
breaches that occur. The same pattern is repeated for the discussion of automobile
accidents, while the development of each of the other topics nuisance law, pollution
control, law enforcement, product liability, principal-agent liability, litigation and
regulation is contained within a single chapter. ###

6

Chapter 2
Efficiency and Equity

For the purposes of this book, the term efficiency will refer to the relationship
between the aggregate benefits of a situation and the aggregate costs of the situation; 4
the term equity will refer to the distribution of income among individuals. 5 In other
words, efficiency corresponds to “the size of the pie,” while equity has to do with
how it is sliced. Economists traditionally concentrate on how to maximize the size of
the pie, leaving to others__ such as legislators__the decision how to divide it. The
attractiveness of efficiency as a goal is that, under some circumstances described
bellow, everyone can be made better off is society is organized in and efficient
manner.
Is There a Conflict?
An important question is whether there is a conflict between the pursuit of
efficiency and the pursuit of equity. If
4.This popular concept of efficiency is more intuitive than the technical
concept of efficiency known as Pareto efficiency or Pareto Optimality (after the
Italian Economist Vilfredo Pareto). A situation is said to be Pareto efficient or Pareto
optimal if there is no change from that situation that can make someone better off
without making someone else worse off. Equivalently, if a situation is not efficient in
this sense, then, by definition, someone can be made better off without making
anyone else worse off. I have opted for the money intuitive concept of efficiency used
in the text for expositional simplicity.
5.This is the standard sense in which economists use the term equity. However,
lawyers and philosophers often use this term differently. For example, equity might
refer to the process by which income or wealth is acquired (as opposed to its final
distribution), or to the degree to which exogenously determined rights are protected.
the pie can be sliced in any way desired, then clearly there is no conflict__with
a bigger pie, its division must be quite unequal, then, depending on what constitutes
an equitable division of the pie, its division must be quite unequal, then may well be a
conflict between efficiency and equity. It may be preferable to accept a smaller pie
(less efficiency) in return foe a fairer division (more equity).

7

The potential conflict between efficiency and equity an be illustrated by a
fanciful example. Suppose that the government must decide whether to build a dam
and that the dean of Stanford law school and I are the only two individuals affected
by it. Currently without the dam, the Dean has \$65 and I have \$35, so the total sum is
\$100. The dam would cost \$30 worth of my labor but none of the Dean's. The dam
would create benefits worth \$40, all of which would go to the Dean because the only
feasible location for building the dam happens to be on her property. Should the dam
be built?
On efficiency grounds, the dam clearly it should be built because it creates
benefits \$40 and costs only \$40 and costs only \$30, thereby creating net benefits of
\$10. But the equity effects need to be considered as well. Before the dam is built, the
Dean has \$65 and I have \$35. After the dam is built, the dean will have \$105
(including the \$40 benefit) and I will have \$5 (after subtracting my \$30 cost).
Whether these distributional consequences are desirable depends on what constitutes
a fair distribution of income. Suppose that the most equitable distribution of income
involves the Dean receiving 60 percent of the total income and my receiving 40
percent. If the dam is not built, then the Dean should have \$60 and I should have \$40.
If the dam is built the total income raises by \$10, the Dean should have \$66 and I
should have \$44.
But suppose, regardless of whether the dam is built, it is impossible to
redistribute income between the two of us. Therefore, The choice is between the
Dean's having \$65 and my having \$35 if the dam is not built, and the Dean's having
\$105 and my having \$5 if the dam is built. Building the dam is more efficient but less
equitable. How this conflict between efficiency and equity should be resolved
depends on how important, it might be desirable to sacrifice some efficiency for more
equity by not building the dam (in other words, “damn” the Dean).
Aternatively, suppose that ir is possivle to costlessly redistribute income
between theDEan and me. Then, given the preferred distribution of income, if the
dam is not built, \$5 would be transferred from the Dean to me, so that she would end
up with \$60 dollars and I would have \$40 dollars. If the dam is built \$39 dollars
would be transferred to me, so that she would have \$66 dollars and I would have \$44
dollars. Clearly, since total income is distributed according to the percentage desired
and we both are better off with the dam, the dam should be built again, there is no
conflict between efficiency and equity.
Note that, if it is possible to redistribute income at no cost, the dam should be built
regardless of what constitutes an equitable distribution of income. If, for example, an
egalitarian income distribution is desired, then without the dam the Dean and I each
could have \$50 dollars and with the dam we each could have \$55 dollars. If
alternatively, equity required that everything should go to the Dean, the dam should
be built because she then could have \$110 dollars rather than \$100 dollars.
The dam example illustrated two important general observations.
If income cannot be costlessly redistributed, there may be a conflict between

8

efficiency and equity. Whether there is in fact a conflict depends on the specific
distributional consequences of pursuing efficiency and on what constitutes an
equitable distribution of income. However, if income can be costlessly redistributed,
there is no conflict between efficiency and equity. This is true regardless of the
distributional consequences of pursuing efficiency and regardless of what constitutes
and equitable distribution of income. I other words, if income can be costlessly
redistributed, if it always preferable to maximize the size of the pie because the pie
can be sliced in any way desired.
Whether the income can be costlessly redistributed is discussed in Chapter 18.
Although the “conclusion” there is that income redistributions generally is costly, it is
argued, nonetheless that efficiency should be the principal criterion for evaluating the
legal system. The argument rests on the observations, explained at length in Chapter
18, that it is often impossible to redistribute income through the choice of legal rules
and that, even when it is possible, redistribution through the government’s tax and
transfer system may be cheaper and is likely to be more precise. In other words, the
potential conflict between efficiency and equity when income redistribution is costly
should be considered in the design of the government’s tax and transfer system, but
not generally in the choice of legal rules. Thus, for purposes of discussing the legal
system, a reasonable simplifying assumption is that income can be costlessly
redistributed. This assumption will be maintained until Chapter 18 (although the
distributional consequences of legal rules occasionally will be noted).
11

Before proceeding, it is worth mentioning several other standard assumptions of
economic analysis
that will be made in analyzing the efficiency of legal rules. First, all benefits end cost
can be measured in terms of a common denominator-----dollars. This assumption is
made for expositional simplicity. It is not essential to economic analysis and it does
not include consideration that might be thought as non economic-----such as the
protection of life and limb.*6 Second, individuals themselves determine the dollar
values to places on their benefits and costs. This is referred to as the assumption of
consumer sovereignty. It is an acceptable assumption if one believes that individuals
generally know what is best for themselves. Third, the values that individuals place
on their benefits and costs are “stable” in the sense that these values are not affected
by changes in public policy. For example an individual’s evaluation of the desirability
of cleaner air is assumed not to depend on whether the legal system establishes the
right to clean the air. Finally, individuals (and, when relevant, firms) maximize their
benefits less their costs. This is known as the assumption of utility maximization or,
when firms are involved, profit maximization.

9

*6. However, to incorporate benefits and costs that are not equivalent to a gain or
loss of money would require the introduction of economic concepts that are beyond
the scope of this book.
13
CHAPTER 3
THE COASE THEOREM
One of the central ideas in the economic analysis of law was developed in an article
by Ronald H. Coase in 1960. *7
This idea, which has since been named the Coase Theorem, is most easily described
by an example.
Consider a factory whose smoke causes damage to the laundry hung outdoors by five
nearby residents. In the absence of any corrective action each resident would suffer
\$75 in damages, a total of \$375. The smoke damage can be eliminated in either two
ways: a smoke screen can be instolled on the factory’s chimney, at a cost of \$150, or
each resident can be provided with an electric dryer, at a cost of \$50 per resident. The
efficient solution is clearly to install the smokescreen because it eliminates total
damages of \$375 for an outlay of only \$150, and it is cheaper than purchasing five
dryers for \$250.
Zero Transaction Costs
The question asked by Coase was whether the efficient outcome outcome would
result in the right to clean air is assigned to the residents or if the right to pollute is
given to the factory. If there is a right to clean air, then the factory has three choices:
pollute and pay \$375 in damages, install a smokescreen for \$150, or purchase five
dryers for the residents at a total cost of \$250. Clearly, the factory would install the
smokescreen, the efficient solution. If there is a right to pollute, then the residents
face three choices: suffer their collective damages of \$375, purchase five dryers for
\$250, or buy a smokescreen for the factory for \$150. The residents also would
purchase the smokescreen. In other words, the efficient outcome will be achieved
regardless of the assignment of the legal right.
It was implicity assumed in the preceding paragraph that the residents could
costlessly get together and negotiate with the factory. In Coase’s language, this is
referred to as the assumption of zero transaction costs. In general, transaction costs
include the costs of identifying the parties with whom one has to bargain, the cost of
getting together with them the cost of the bargaining process itself, and the costs of
enforcing any bargain reached. With this general definition of transaction costs in
mind, we can now state the simple version of the Coase Theorem: If there are zero
transaction costs, the efficient outcome will occur regardless of the choice of legal
rule.
Note that, although the choice of the legal rule does not affect the attainment of the

10

efficient solution when there are zero transaction costs, it does affect the distribution
of income. If the residents have the right to clean air, the factory pays \$150 for the
smokescreen, whereas if the factory has the right to pollute, the residents pay for the
smokescreen. Thus, the choice of the legal rule redistributes income by the amount of
the least-cost solution to the conflict. Because it is assumed tributional effect of no
consequences.---if it is not desired, it can be easily corrected.
*7 Ronald H. Coase, the problem of social Cost, 3 J.L. &Econ. 1(1960)
Positive Transaction Costs
The assumption of zero transaction costs obviously is unrealistic in many conflict
situations. At the very least, the deputing parties usually would have to spend time
and/or money to get together to discuss the dispute. To see the consequences of
positive transaction costs, suppose in the example that it costs to each resident \$60 to
get together with each other (do, say, transportation costs and the value of each
resident’s time).

15
If the residents have a right of clean air, the factory again faces the choice of paying
damages, buing a smokescreen, or buying 5 dryers. The factory again would
purchase the smokescreen, the efficient solution. If the factory has the right to
pollute, each resident now has to decide whether to bear the loss of \$75, buy a dryer
for \$50, or get together with other residents for \$60 to collectively buy a smokescreen
for a \$150. Clearly, each resident will choose to purchase a dryer, an inefficient
outcome. Thus given the transaction costs described, the right to clean air is efficient,
but the right to pollute is not.
Note that in the example the preferred legal rule minimized the effects of
transaction costs in the following senses. Under the right to clean air, the factory had
to decide whether to pay damages, install a smokescreen, or buy 5 dryers. Because it
was not necessary for the factory to get together with the residents to decide what to
do, the transaction costs---the costof te residents to get together---did not have any
effect. Under the right to pollute, the residents had to decide what to do. Because
the residents were included to choose an inefficient solution in order to avoid the
costs of getting together, the transaction costs did have an effect. Thus, even though
no transaction costs were actually incurred under the right to pollute because the
residents did not get together, the effect of transaction costs were greater under that
rule.
We can now state the more complicated version of the Coase Theorem: If there
are positive transaction costs, the efficient outcome may not occur under every legal
rule. In these circumstances, the preferred legal rule is the rule that minimizes the
effect of transaction costs. These effects include actually incurring transaction costs

11

as well as the inefficient choices induced by a desire to avoid transaction costs.
The distributional consequences of legal rules are somewhat more complicated
when there are transaction costs. It is no longer true, as it was when there were zero
transaction costs, that the choice of the rule redistributes income by the amount of the
least-cost solution. In the example, if the residents have the right to clean air, the
factory pays \$150 for the smokescreen, whereas, if the factory has the right to pollute,
the residents pay \$250 for 5 dryers.
Although the simple version of the Coase theorem makes an unrealistic
assumption about transaction costs, it provides a useful way to begin thinking about
legal problems because it uggests the kinds of transaction that would have to occur
under each legal rule in order for that rule to be efficient. Once these required
transactions are identified, it may be apparent that, given more realistic assumptions
about transaction costs, one rule clearly is preferable to another on efficiency ground.
The more complicated version of the Coase Theorem, provides a guide to choosing
legal rules in this situation. All of the application investigated in this book--including nuisance law, breach of contract, automobile accident, pollution control,
products liability, and litigation---can be approached in this way, although some fit
more naturally into the Coasian framework than others.

12

CHAPTER 4
FIRST APPLICATIONNUISANCE LAW
One area of law that can be readily examined in terms of Coase Theorem is
nuisance law.
Nuisance cases result from incompatible land uses and typically involved a small
number of individuals bargaining with each other, as when emissions from a factory
fall upon neighboring property, bright lights or noise disturb someone’s sleep, or an
unsightly building spoils an attractive residential neighborhood.
Adopting a framework first suggested by Guido Calabresi and A. Douglas
Melamed, *8
The resolution of a nuisance dispute may be viewed as involving two steps. First, an
entitlement
must be chosen---that is, a determination must be made regarding who is entitled to
prevail. The injurer can be granted the right to engage in the activity that causes
harm, or the victim can be granted the right to be free from harm. Then, a decision
must be made about how to protect the entitlement. One possibility is to grant the
holder of the entitlement an injunction. If the victim hold the entitlement, protecting
it by an injunction means that she can prohibit the injurer from
causing harm. Thus, the injurer can cause damage only if he “buys off” the victim.
Similarly, if the injurer holds the entitlement, protecting it by an injunction means
that the victim must buy ff the injurer if she wants damages reduced.

*8 Guido Calabresi &A. douglas Melamed, Property rules, Liability Rules and
Inalienability: One view of the Cathedral, 85 Harv. L. Rev. 1089 (1972).

13

18
An alternative method of protecting entitlements is to give the holder of the
entitlement an amount of money---damages---that some governmental body, such as
a court, determines. If the victim has the entitlement, she has the right to be
compensated, but she cannot prohibit the injurer from causing harm, as she could
under an injunctive remedy. *9
Analogously, if the injurer holds the entitlement, protecting him by a damage remedy
would mean that the victim could restrict the injurer’s activity but would have to
compensate the injurer for the injurer’s “damages”(for example, forgone profit). This
last combination—entitling the injurer to damages—is unconventional, but it has
been used. *10 Thus, there are four possible solutions, corresponding to who is
given the entitlement and how it is protected.
In this chapter, we will examine whether the efficiency criterion can determine
which entitlement to choose and which remedy to use to protect it. The analysis will
be based on an example of a polluting factory located next to a single resident. The
example is described in
Table 1. The factory can produce zero, one, two, or three units of output. Increasing
output results in additional profits for the factory and additional damages to the
residents. If the factory produces one unit, the factory obtain \$10,000 in profits and
the resident suffer \$1,000 in damages. Total profits less total damages are \$9,000. If
the factory produces a second unit, the factory’s additional profits are \$4,000---so the
factory’s total profits are \$14,000—and the resident’s additional damages
are \$15,000—so the resident’s total damages are \$16,000. Then total profits less total
damages are -- \$2,000. If the factory produces a third unit, the results are described
similarly.
Maximizing the size of the pie in the nuisance law example is equivalent to
maximizing the factory profit net of the resident’s damages. Given the data in table
1, this occurs when the factory produces one unit of output (see the last column). In
other words, one unit of output is the efficient solution.

14

TABLE 1
Nuisance Law Example
Factory
Profits of Profits
Factory
factory

Total
of Damages of Damages of Profits Less
Resident
Residents
Total
Damages
--\$0
\$0

0

---

\$0

1

\$10,000

\$10,000

\$1,000

\$1,000

\$9,000

2

\$4,000

\$14,000

\$15,000

\$16,000

-\$2,000

3

\$2,000

\$16,000

\$20,000

\$36,000

-\$20,000

*9 A leading American nuisance case illustrating this version of the damage remedy
in Boomer v. Atlantic Cement Co., 26 N.Y. S. 2d 219, 257 N.E. 2d 870, 309 N.Y. S.
2d 312 (1970).
*10 See your Industries, Inc. v. Del E. Webb Development Co., 108 Ariz. 178, 494
P.2d 700(1972)

15

19
Zero Transaction Costs
We know from the discussion in the previous chapter that if there are zero
transaction costs, then the factory will end up producing the efficient output
regardless of the choice of remedy or entitlement. it will be useful to see how this
costs.
Under the injunctive remedy, suppose, for example, that an entitlement to clean
air is given to the residents. This corresponds to giving the resident the right to force
the factory to produce zero output. However, the factory, would gain \$10,000 in
profits from producing one unit, and the resident would suffer only \$1,000 in
damages. Thus, it is in each party’s interest to reach an agreement in which the
factory would pay the resident some amount between \$1,000 and \$10,000 for
permission to produce one unit. Assuming zero transaction costs, which is
interpreted to imply cooperative behavior, such an agreement will be reached. It will
not be mutually beneficial for the factory to produce a second unit because the
factory would gain only an additional \$4,000 in profits, whereas the resident would
suffer an additional \$15,000 in damages. Similarly, it would not be mutually
beneficial for the factory to produce three units. Thus, the parties would remain at
one unit, the efficient solution.
Under the damage remedy, suppose that an entitlement to clean air is given to
the resident, as before, and that the court makes the factory liable for the resident’s
actual damage. Because the factory would gain \$10,000 from producing one unit and
would be liable for only \$1,000, the factory clearly will choose to produce at least
one unit. It will not be in the factory’s interest to produce the second unit, however,
because the increase in the factory’s profits is \$4,000 and its additional liability is
\$15,000. Similarly, the factory would be worse off if it produced three units. Thus,
the factory would choose to produce the efficient output.
Strategic Behavior

16

The assumption of zero transaction costs obviously is unrealistic in many
respects. We will first consider the possibility that the parties will behave
strategically;*11 for instance, to establish a reputation as a tough bargainer, a party
may “hold out” for a disproportionate share of the gains from any agreement. If both
parties are stubborn in this sense, they may not be able to reach an agreement even
when bth could be made better off. Such Behavior is not un common; for example,
parties often go to court rather than settle out of court more cheaply.*12 We will now
consider the nuisance remedies assuming that the parties behave strategically.
*11. Although strategic behavior does not necessarily generate any out-ofpocket cost or costs associated with lost time, it will be treated as a type of
transaction cost. It is like other transaction costs in that it may prevent the parties
from reaching an efficient agreement.
*12. There may be reasons other than strategic behavior why parties litigate
rather than settle. For example, the parties may disagree about the plaintiff’s chance
of winning at trial and therefore may not perceive a mutually beneficial settlement.
See Chapter 16 below.
21
Under the injunctive remedy with an entitlement to clean air, we saw that it
would be in each party’s interest t reach an agreement in which the factory paid the
resident some amount between \$1,000 and \$10,000. However, because of strategic
behavior, the resident may for example, hold out for \$8,000, while the factory may
refuse to pay anything over \$5,000. As a result the resident might enforce the
injunction and shut down the factory ( at least for some period), an inefficient
outcome..
The problem of strategic behavior under the injunctive remedy can be
overcome by the appropriate choice of the entitlement. Instead of an entitlement
corresponding to zero output of the factory—an absolute entitlement to clean air—
suppose that the court were to choose an entitlement
Corresponding to the unit of output—an intermediate entitlement. Under the
injunctive remedy, this would mean that the factory could produce one unit, but no
more, without having to obtain the permission of the resident. Starting at one unit of
output, it would not be mutually beneficial to produce a second or third unit because
the factory’s gains are less than the resident’s losses. Likewise, it would be not
mutually beneficial to reduce output to zero because the resident’s gain
(in the form of reduced damages) is less than the factory’s loss ( in the form of
reduced profits). Thus, starting at an intermediate entitlement of one unit, the parties
will remain there. Strategic behavior cannot upset this outcome because no beneficial
changes can be made that would require negotiation.
This discussion illustrates a general principle: under the injunctive remedy, to
overcome strategic behavior it is necessary to choose an entitlement corresponding to
the efficient outcome.

17

This is because, starting from any other entitlement, the parties must reach an
agreement to get to the efficient outcome; and strategic behavior may prevent this
agreement from being reached.
Under the damage remedy with an entitlement to clean air and liability equal to
actual damages, we saw that the factory would choose to produce the efficient output
of one unit. The presence of strategic behavior does not affect this result because
there are no bargains that the parties have to reach. Also, there are no threats that the
factory can make because given that the factory is liable for actual damages, the
resident is indifferent among all levels of the factory’s output.
In the essential to the conclusion of the previous paragraph that liability equals
actual damages. To see why, suppose that the factory liability is \$700 for the first unit
—exceeding the resident’s damages of \$1,000—and as before, is equal to the
resident’s damages for the second and third units. If the factory produces one unit of
output it will gain \$3,000 (\$10,000-\$7,000). The resident also will gain \$6,000, the
amount by which the liability payment exceeds actual damages (\$7,000--\$10,000).
But the factory can deny this gain to the resident by not producing the first unit of
output. Therefore, if the factory believes that it can bargain more effectively than the
resident, it might threaten to not produce unless the resident pays some specific
amount up to her full gain or \$6,000. However, if the resident believes that she is the
better bargainer, she might not give in to the factory’s demand. As a result, the
factory might carry out its threat, if only to make future threat credible and produce at
an inefficient output.
The kind of “extortion” threat just described cannot occur if liability is equal to
actual damages. Because the resident then is not overcompensated, and so gains
nothing from an increase in the factory’s output, the factory’s threat to not produce
the first unit of output, the factory’s threat to not produce the first unit of output has
no effect. With a full compensation, the resident is indifferent to whether the first unit
is produced and is likely indifferent with respect to the second and third units. Thus,
the factory will minimize its, after-liability profits by increasing production to the
efficient output of one unit.
The discussion illustrates another general principle: Under the damage
remedy, to overcome strategic behavior it is necessary to set liability equal to actual
damages. If liability exceeds actual damages, then the party who is liable has on
incentive to threaten to deny the other party’s overcompensation by choosing an
inefficient outcome.*13
The analysis of strategic behavior under the injunctive and damage remedies
suggest another way in which the assumption of zero transaction costs is likely to be
unrealistic. For both remedies, it was seen that the court must have certain
information about the nuisance dispute in order to achieve the efficient solution.
Under the injunctive remedy, the court needs to know the efficient outcome to choose
an entitlement corresponding to it. And under the damage remedy, the court needs to
know the resident’s damages to set liability equal to actual damages. It was implicitly
assumed that the court had whatever information was required. We will now
reconsider the remedies when this information is imperfect. The assumption of

18

strategic behavior will be maintained in this discussion.
Suppose that the court has limited information of the following sort: It knows
the resident’s schedule of damages but does not know the factory’s schedule of
profits. For example, the court might easily be able to obtain information about
damage to property from pollution but not about the cost to the polluter of changing
production methods to abate pollution.
Under the injunctive remedy, the court no longer can achieve the efficient
outcome. To reach that outcome, strategic behavior must be avoided, which requires
under the injunctive remedy that the entitlement coincide with the efficient output.
But to determine the efficient output, the court must know when the factory’s profits
net of the resident’s damages are maximized. Knowing the damage schedule alone
obviously is insufficient to determine this level of output. Although the court could
attempt to estimate the efficient output, if it makes a
*13. More generally, strategic behavior can be overcome under the damage
remedy if liability is less than or equal to actual damages up to the efficient output
and greater than or equal to actual damages beyond the efficient output. It is beyond
the scope of this chapter to explain this more general proposition. Note, however
that the statements and examples in the text are consistent with it.
*14. The alternative case in which the court knows the factory’s profit schedule
but not the resident’s damage schedule will not be analyzed because it is analogous
to the case discussed in the text.
Mistake, as it generally will, strategic behavior might prevent the parties from
bargaining to the efficient output.
The damage remedy can reach the efficient outcome despite the court’s
imperfect information. This result can be guaranteed, however, only if the court gives
an absolute entitlement to clean air to the resident and set liability equal to actual
damages, Any other entitlement might lead to the efficient outcome, but in need not.
For example, suppose that the court assigns an intermediate entitlement to pollute to
the factory liable thereafter for the resident’s damages. Initially, the factory would
choose to produce two units because there is no liability up to and including the
second unit, and liability for the third unit---equal to the resident’s damage of \$20,000
--- exceeds the factory’s additional profits --- equal to \$2,000. The resident then
would have and incentive to “bride” the factory to reduce the output from the two
units to the efficient output of one unit, but because of strategic behavior the parties
might not reach that output.
On the other hand, if the court chooses an entitlement corresponding to zero or
one unit of output and sets liability thereafter equal to actual damages, the damage
remedy will lead the factory to produce at the efficient output. In other words, with
liability equal to actual damages, the damage remedy leads to the efficient outcome
if, and only if, the entitlement is at or below the efficient output. However, because
the court cannot determine the efficient output with its limited information, the only
way it can guarantee the efficient result is to choose the entitlement corresponding to

19

the lowest possible output---an absolute entitlement to the resident.
The discussion thus far has shown that id the court knows the victim’s schedule
of the damages but not the injurer’s schedule of profits, it generally cannot implement
an efficient injunctive remedy but it can implement an efficient damage remedy.
However, in many nuisance situations the court might not be able to easily determine
the victim’s damages. For example, although a court might be able to accurately
estimate the market price of someone’s home, this price generally is less than the
damages that would be suffered by the resident if she were forced to move, because it
does not reflect the special attachment she might have for that location and house.
Damage often includes a subjective or idiosyncratic element of this sort that is
difficult of impossible to measure. 15 We will therefore briefly reconsider the
remedies when the court is assumed to underestimate the resident’s damage (and, as
before, to not know anything about the factory’s profits). For concreteness, suppose
in the example that the court’s estimate of damages is \$500 per unit of output.
Because the court obviously still cannot implement an efficient injunctive
remedy, the discussion will focus on the damage remedy. Suppose an absolute
entitlement is awarded to the resident. If damages were measured accurately, then, as
we saw above, the damage remedy would lead to the efficient outcome. Now,
however, with damages underestimated, the factory generally will overshoot the
efficient output. In the example, with liability equals to \$500 per unit of output, the
factory will chooses to produce three units because its additional profit from
producing each unit exceeds \$500 (see Table 1).
Starting at an output of three units, the resident would be better off by \$19,500
is the output were reduced by one unit---she would lose a \$500 liability payment but
her damages would decline by \$20,000 (see Table 1). The factory would lose only
\$1,500 as a result of this change---its profits would fall by \$2,000, but it would avoid
a \$500 liability payment. Thus, if the parties could reach an agreement in which the
resident paid the factory some amount between \$1,500 and \$19,500 to reduce output
by one unit, both parties would be better off. However, because of strategic behavior,
such a deal will not necessarily occur. And even it an agreement when they negotiate
over reducing output from two units to the efficient output of one unit.
 Foooot Noooote 15
The general point of this discussion can be simplified stated. If courts
underestimate the victim’s damages, then the damage remedy initially will lead
to an excessive output and this inefficiency might not be corrected because of
strategic behavior. There is then no general reason to believe that a damage
remedy would be preferable to an injunctive remedy. For example, suppose
that, starting with an absolute entitlement to the resident, the damage remedy
would lead to an output of zero units because of strategic behavior. The
injunctive outcome is more efficient than the damage outcome because total
profits less total damages are \$0 rather than -\$20,000 (see Table 1). In general,

20

however, either remedy could be the more efficient one.
We can now summarize the results in this chapter regarding the
efficiency analysis of nuisance remedies. If the parties can be expected to
bargain cooperatively (and there are no other transaction costs), then any
choice of entitlement and remedy will be efficient outcome still can be
achieved under both remedies if the court has adequate information. Strategic
behavior can be overcome under the injunctive remedy by choosing the
entitlement that corresponds to the efficient outcome, which can be determined
if the court knows the injurer’s benefits from engaging in the harmful activity
and the victim’s damages. And strategic behavior can be overcome under the
damage remedy by giving an absolute entitlement to the victim and setting
liability equal to actual damages, which obviously requires knowledge of the
victim’s damages. If the court only knows the victim’s damages, the injunctive
remedy generally will fail because the court cannot accurately set the
entitlement equal to the efficient outcome, but the damage remedy still can
guarantee the efficient outcome. However, if the court underestimates the
victim’s damages, then te damage remedy generally will lead to excessive
output and may be less desirable than the injunctive remedy.
Although in theory either remedy could therefore be more efficient, it
may be apparent in particular circumstances that one remedy is likely to be
better than the other. For example, suppose that a court is confident that its
estimate of the victim’s damages is close to the truth, but believes that its
estimate of the injurer’s benefits is inaccurate. Then an entitlement to the
victim protected by a damage remedy generally would be preferred because
this would be likely to lead to an outcome close to the efficient solution.
Alternatively, suppose that a court has poor information both about the victim’s
damages and the injurer’s benefit, but is confident that the efficiency loss from
too little activities by the injurer is small relative to the efficiency loss form
excessive activity. Then an entitlement to the victim protected by an injunctive
remedy would be desirable because this would guarantee that the final outcome
will not be too bad. Thus, the efficiency analysis of nuisance law may be
remedy to choose.

21

29
CHAPTER 5
SECOND APPLICATION—
BRECH OF CONTRACT
Another area of law that can be discussed within the Coasian framework of
bargaining among a small number of individuals is contract law. Unlike the
normal nuisance law situation, however, the parties to a contract negotiate with
each other before any dispute arises. Since the parties can decide in advance
how to resolve potential disputes, it might be asked whether it is necessary or
desirable to have general legal rules that govern contract disputes. The reason
contract rules are desirable is, of course, that it would be prohibitively costly,
(if even possible) to negotiate and draft a contract that provides for every
conceivable contingency. For contingencies that are thought to be unlikely or
that do not affect the parties’ costs and benefits very much, it is not worth
going to the trouble to specify in advance what to do if the contingency should
occur.
Contract law can be viewed as filling in these “gabs” in the contract—
attempting to reproduce what the parties would have agreed to if they could
have costlessly planned for the event initially. Since the parties would have
included contract terms that maximize their joint benefits net of their joint costs
—both parties can thereby be made better off—this approach is equivalent to
designing contract law according to the efficiency criterion.*16
In this chapter we will examine three remedies for breach of contract

22

from this perspective. One, expectation damages, awards the breached-against
party an amount of money that puts her in the same position she would have
been in had the contract been completed. Another, reliance damages, awards
an amount of money that places the breached-against party in the position she
would have been in had she never entered into the contract initially. The last,
restitution damages, awards the breached-against party an amount of money
corresponding to any benefits that she has conferred upon the breaching
party.*17
The analysis will be undertaken using an example in which a seller, S,
can produce a good called “widget” for %150. Widgets are not generally
available. For this reason a
buyer, B1, who values the widget at \$200 ,enters into a contract with S for the
future deleivery of a widget. The contract price is paid in advance. In order to
use the widget, B1, must make an expenditure of \$10 prior to delivery (for
example, she might have to modify her warehouse or store the widget). This
will be referred to as her reliance expenditure or reliance investment. If the
contract is not completed, this expenditure is assumed to have
no value.*18
Before delivery occurs there is the chance that some other buyer, B2,
may also want the widget. The value B2 attaches to the widget in not known at
the time S and B1 enter into their contract. For simplicity, it is assumed that
B2’s value will be either \$0, \$180, or \$250 and that he will offer this amount
for it. Thus, after S and B1 have entered into their contract, there is a chance
that B2 will offer more than B1did. Both S and B1are assumed
to be aware of this possibility. The fact of this example are summarized in
Table 2 (ignore for now the note at the bottom of the table).
*16. The statement that the parties would have maximized their joint benefits
net of their joint costs obviously presumes that they would have bargained
cooperatively. Also the conclusion that the maximization of the parties’ joint
benefits net of their joint costs in the goal of efficiency presumes that no one
else is affected by the contract.

23

31
TABLE 2
BREACH OF CONTRACT EXAMPLE
S is the seller:
S’s costs to produce the widget is \$150.
B1 values the widget at \$200.
B1’s reliance expenditure is of \$10.*a
B1 pays S the contract price in advance.
B2 values the widget at \$0, \$180, or \$250.
*a. The possibility that B1can spend an additional \$24 on reliance and thereby
increase the vaue of the widget to her by \$30 also is considered.
It also will be assumed in this chapter that the parties are risk neutral. This
means that they only care about the expected value of a risky situation—that is, the
magnitude of a potential loss or gain multiplied by the probability of the loss or gain
occurring. For example, the espected gain in a situation involving a 50 percent
chance of winning \$10,000 is \$5,000. A risk –neutral person would, by definition, be
indifferent between this situation and any other one with the same expected gain—
such as a situation involving a 25% chance of winning \$20,000, or one involving a
certainty of winning \$5,000.

24

32
A fully Specified Contract
Before considering how breaches of contract remedies fill in the gaps in
incompletely specified contracts, it will be useful to examine the contract between S
and B1 when everything is specified in advance. Suppose the contract has the
following provisions.
First, the contract price, payable in advance, is \$175. Second, if the value of
B2 attaches to the value to the widget is \$0 or \$180, then S is to deliver the widget to
B1. In this case, s’s profit is \$25—the contract price of \$175 less S’s production cost
of \$150—and B1’s profit is \$15—the \$200 value of the widget to B1 less her \$10
reliance expenditure and less the contract price of \$175. Third the contract states that
in the event that B2 values the widget at \$250, S is to sell it to B2 rather than B1, but
then must pay B1 \$225. In this case S’s profit is \$50—the contract price of \$175 less
S’s production cost of \$150, plus B2’s payment to S of \$250 less S’s payment to B1of
\$225. B1’s profit is then \$40—the \$225 payment form S less the reliance expenditure
of \$10 and less the contract price paid in advance of \$175.
It is clear that this full specified contract between S and B1is efficient. The
only choice that parties have to make that affects their joint profits is whether S is to
sell to B2 if B2 wants the widget. If S does not sell to B2, then the joint profits of S
and B1 are \$40—S’s profits of \$25 plus B1 profits of \$15. If S does sell to B2 when
B2 values the widget at \$250 then the joint profit of S and B1 raise to \$90—S’s profit

25

of \$50 plus B1’s profit of \$40. However, if S were to sell to B2 when B2 values the
widget at \$180, the joint profits of S and B1 would fall to \$20 because, together, they
would have revenue of \$180 and costs of \$160 (S’s production cost of \$150 plus B1
reliance expenditure of \$10). Thus, the provisions in the contract that call for S to sell
the widget to B2
if B2 values it at \$250, but if he values it at \$0 or \$180, are efficient.*19
It is important to note that there is a close relationship between the contract
price and the amount of money S has to pay to B1 if S sells the widget to B2. In
general, the higher the amount paid to B1 the higher the contract price. For example,
suppose S has to pay B1 \$240 rather than \$225 in the event that the widget is sold to
B2. B1 clearly would prefer to receive the higher payment in the event of a breach,
and S clearly would prefer to pay the lower amount. Presumably, therefor, S will
demand, and B1 will be willing to offer a higher contract price in advance—say \$180

*17. Another remedy, liquidated damages, will be discussed in Chapter 8
below.
*18. The assumptions that the contract price is paid in advance and that the
reliance investment has no value in the event of breach are not essential and do
not affect an of the general conclusions in this chapter.
*19. Since B2 is assumed to offer an amount equal to the value he attaches to the
widget, his profits are not affected by how valuable the widget is to him. This is why
B2’s profits do not need to be taken into account by determining the efficient contract
provisions.
33
Efficient Breach
Thus far, the example illustrates a simple, but fundamental principle in the
economic analysis of contract law: A fully specified contract is efficient. In practice,
however the cost of contracting will lead the parties to ignore relatively unimportant
contingencies. Therefore, now suppose that because S and B1believe that an offer
from B2 is unlikely, they do not bother to include a provision in the contract that
deals with the possibility that B2 will offer more for the widget.

26

The contract simply states that S is to deliver a widget to B1 at some price payable in
advance. We will now examine whether the expectation, reliance, and restitution
remedies for breach of contract are efficient alternative to an explicit contract
provision regarding when the widget should be sold to B2 rather than to B1. An
important assumption is the following analysis is that if S wants to breach the
contract, B1 will not find it worthwhile (because of bargaining costs) to attempt to
stop S from breaching or to purchase the widget from B2 after the breach. (If it were
costless for B1 to negotiate with S or B2, than the Coasian analysis in Chapter 3
implies that every remedy would be efficient.)
First consider a breach of the contract when the expectation remedy is
applicable. If the contract had been completed, B1 would have made a profit equal to
the \$200 value she places on the widget less her reliance expenditure and less the
contract price she paid in advance. Thus, to put B1 in the same position she would
have been in had the contract been completed, it is necessary to compensate B1 \$200
(because B1 has already incurred the reliance expense and has paid S the contract
price). Given a damage payment of \$200, S will decide to breach if B2 offers \$250 for
the widget, but not if he offers \$180(or, of course, \$0). Thus, the expectation remedy
leads to the efficient outcome. Put differently the expectation remedy is an efficient
substitute for the explicit contract provisions governing when S is to sell the widget to
B1 or B2. It thereby saves the parties the cost and inconvenience of dealing with
unlikely contingencies every time they enter into contract. Instead, they can simply
rely on a breach of contract remedy in the few instances when the contingency arises.
Note that the conclusion that the expectation remedy includes efficient breach
decisions does not depend on what the contract price is. S will have to pay B1 \$200 in
the event of a breach regardless of the contract price because, given that B1 paid the
contract price( and incurred the reliance expense) in advance, \$200 is the amount of
money necessary to put B1 in the same position she would have been in had the
contract been completed. Thus, regardless of the contract price, S will breach in order
to sell to B2 only if B2 offers more than \$200 for the widget.
Next consider the reliance remedy. If B1 had not entered into a contract with S,
it is assumed that she would have earned zero profit. 20 thus, to put B1 in the same
position she would have been in had she never entered into contract, it is necessary to
compensate B1 for her \$10 reliance investment and to return to her the contract price
that she paid in advance. In other words, the reliance measure of damages equals the
reliance expenditure pule the contract prince.
To determine the effects of the reliance remedy on S’s decision to breach, it is
therefore necessary to examine the setting of the contract price. Since S’s production
cost is \$150, he will not be willing to accept less than this amount. B1 will be willing
to pay up to \$190 for the widget because she values it at \$200 but has to make the
\$10 reliance investment. Thus, the contract price will be somewhere between \$150
and \$190, the exact price depending on the relative bargaining strengths of the
parties. Suppose, for concreteness, that is it \$160. Given a reliance remedy
expenditure of \$10 and a contract price of \$160, the reliance remedy would award B1
\$170 in the event of a breach by S. Therefore, S will breach the contract if B2 offers

27

*20. This assumption is not essential and does not affect any of the general
either \$180 or \$250 of the widget because S has to pay only \$170 in damages. If B2’s
value is \$180, the breach will be inefficient because the value B2 attaches to the
widget is less than the \$200 value B1 attaches to it. In other words, the reliance
remedy might lead to an inefficient breach.
Finally, consider the restitution remedy. The only benefit B1 has conferred on
Sis the contract price that she paid in advance. Thus, to award B1 an amount of
money corresponding to the benefit she has conferred, it is necessary to force S to
return the contract price. For the same reasons discussed with respect to the reliance
remedy, the contract price will be between \$150 and \$190. If it is below \$180, then
the restitution remedy also will lead to an inefficient breach when B2 values the
widget at \$180.
In general, the restitution remedy is more likely to lead to inefficient breaches
than the reliance remedy for the following reasons. First note that the contract price
under the restitution remedy tends to be less than the contract price under the reliance
remedy. Intuitively, this is because the seller (S in the example) does not have to
compensate the buyer (B1 in the example) for her reliance expenditure under the
reliance remedy, but he does have to under the reliance remedy.
Consequently, the buyer presumably would not be willing to pay as much for the
contract under the restitution remedy, the damage payment also would be lower under
that remedy because restitution damages equal contract price plus the reliance
expenditure. One would expect inefficient breaches to be more likely under the
restitution remedy because of the lower damage payment.
The discussion thus far illustrates several general conclusions about the
economic effects of contract law. The key result is that the expectation remedy is the
only remedy that creates efficient incentives with respect to breaches of contracts.
This is because the expectation remedy forces the breaching party to pay damages
equal to the value of the good to the breached-against party. If another buyer values
the good more than this, then it is efficient for the buyer to have the goo. Given the
expectation measure of the damages, the seller will have an incentive to breach in
order to obtain the higher offer. If another buyer values the good less that the original
buyer, a breach in not efficient and the expectation remedy will appropriately
discourage breaches. Any other measures of damages exceed expectation damages,
then a breach might not occur even though it would be efficient. For example, if
damages where 4260 in the example, then S would not breach when B2 offers \$250.
And if the damages are less than expectation damages, and inefficient breach might
occur. This is the problem with the reliance remedy, because it leads to a level of
damages below the expectation level. The restitution remedy is worse because it
results in even lower damages than those under the reliance measure.

28

Efficient Reliance
Including optimal breaches of contracts is not the only problem with which
contract law has to deal. Another issue of concern relates to reliance expenditures. In
the example, it was assumed that B1’s Reliance investment was fixed at \$10. In
general, this expenditure can vary, and the more spent on reliance, the more valuable
the contract will be to the buyer if it is completed. For example, the buyer might be
able to purchase various customized pieces of equipment, each of which is capable of
transforming the widget into a more valuable final product. (Because widgets are
perishable, this equipment must be obtained before delivery.) In the remainder of this
chapter we will analyze how remedies for breach of contract affect the amount
invested in reliance.
In order to examine the reliance decision, the example used earlier needs to be
made slightly more complicated. We will continue to assume that the original buyer
B1 must spend at least \$10 on reliance and that the widget will be worth \$200 to her
if this is all the she spends. But now she will have the option of spending an
additional 424 on reliance and thereby raising the value of the widget to her by \$30.
As before, if the contract is not completed, the reliance expenditure will have no
value. It also will be assumed that the three values the second buyer B2 might attach
to the widget--\$0, \$180, or \$250--are all equally likely. The relevance of this
assumption will become apparent shortly.
Although it might seem that the efficiency criterion would dictate having B1
make the additional reliance investment—because this investment seems to increase
the value of the widget by \$30 at a cost of only \$24—this conclusion is incorrect.
The increase in value occurs only if the contract is completed and B1 obtains the
widget. But as we have seen, it is efficient for S to breach the contract with B1 in
order to sell to B2 if the value of the widget to B2 turns out to be \$250. Given the
assumption at the end of the previous paragraph, there is in one-in-three chance that
B2 will value the widget this much. In other words, if the breach decision is efficient
there is only a two-thirds chance that B1 will obtain the widget. Thus, while the \$24
cost of reliance is certain, the \$30 benefit from reliance is uncertain. The expected
benefit—the benefit multiplied by the probability of its realization—is only \$20. It is
inefficient to incur a \$24 cost to obtain a \$20 expected benefit.
Id S and B1 had been able to costlessly negotiate and draft their contract
initially; they would have included a provision that specified that B1 is not to make
the additional \$24 reliance investment. However, negotiation over b1’s reliance
decision is not a simple matter. For example, if it is difficult for S to verify how much
would obtain from the additional reliance expenditure, B1 might be able to take
advantage of S’s imperfect information. Thus, analogously to the breach of contract
decision, it is reasonable to suppose that a provision regarding the reliance decision
was not included in the contract. We will therefore examine whether some breach of
contract remedy can serve as a substitute for the missing provision.
Under the expectation remedy, B1 either will receive the widget (if the contract

29

is performed) or be given an amount of money equivalent to the value of the widget
(if the contract is breached). If B1 spends \$10 on reliance, the widget will be worth
an additional \$30 to her, so expectation damages would equal \$200. If she spends an
additional \$24 on reliance, the widget will be worth an additional \$30 to her, so
expectation damages would equal to \$230. Thus, by spending an extra \$24, she will
obtain a \$30 benefit either because the widget will be delivered or higher expectation
damages will be paid. Clearly, B1 will invest the additional \$24 in reliance, an
inefficient outcome for the reasons discussed above. In other words, because the
expectation remedy in effect guarantees performance, it does not force B1 to take into
account the fact that the reliance expenditure will be worthless if the contract is
breached. It therefore encourages excessive reliance investments.
Under the reliance remedy, B1 either will receive the widget or be given an
amount of money equal to her reliance expenditure plus the contract price. Thus, if
the contract is performed, the extra \$24 investment in reliance will have been
worthwhile because it will have raised the value of the widget by \$30. If the contract
is breached, the extra \$24 will be returned on reliance because reliance damages will
be that much higher. B1 therefore will have an incentive to spend the extra \$24 on
reliance because this expenditure is, in effect, an investment that has no “downside”
risk but that does have “upside” potential. In other words, the reliance remedy also
encourages excessive reliance expenditures.
Under the restitution remedy, B1either will receive the widget or be given an
amount of money equal to the contract price. Unlike under the expectation remedy,
she is not effectively guaranteed performance, and unlike under the reliance remedy,
she does not get her reliance investment back in the event of breach. B1’s reliance
expenditure is now, in effect, a risky
investment with a positive payoff in the event of performance and no payoff in the
event of breach. Therefore, in order for B1 to determine whether is worthwhile to
spend the extra \$24 on reliance, she needs to know the probabilities of performance
and breach. Suppose S breaches only when it is efficient for a breach to occur—that
is, only when B2 values the widget at \$250.*21
Given the assumption that the three values B2 might attach to the widget--\$0, \$180,
and \$250 are all equally likely, there is a two-third chance of performance and a onethird chance of breach. Thus, the expected benefit to B1 of the reliance expenditure is
\$20—the \$30 increase in value of the widget multiplied by the probability of
obtaining this value. B1 will not spend an extra \$24 in reliance to obtain this expected
benefit. In other words, the restitution remedy leads to the efficient reliance
investment.
This discussion of the effect of breach of contract remedies on the reliance
decision illustrates several general results. The principal one is that, among the
remedies considered, only the restitution remedy induces efficient reliance
investments. It does this because it forces the party investing in reliance to take into
account the fact that the reliance expenditure is worthless id the contract is breached.
*22 the expectation remedy generally leads to too much reliance because it gives the
relying party the value that would have been performed. The reliance remedy also

30

reimburses the relaying party for the cost of reliance in the event of breach.
Another consideration in the economic analysis of breach of contract remedies
is the cost of obtaining the information needed to implement each remedy. The
expectation remedy requires a court to estimate what the value of the contract would
have been to the breached-against party if the contract had been completed. In many
contract situations, this value may be difficult to determine. For example, suppose the
buyer is purchasing specialized memory chips for the production of a newly designed
laptop computer. The court would have to predict how profitable the new computer
would have been. The restitution and reliance remedies both require knowledge of
the contract price, which should be readily available. The reliance remedy also
requires information about the breached-against party’s reliance expenditures. Since
these expenditures will have been made prior to the parties’ coming to the court, it
should be easier for the court to obtain this information than the information required
by the expectation remedy. Thus, in general, the expectation remedy probably would
be the most costly to implement, the restitution remedy would be the cheapest, and
the reliance remedy would be somewhere in-between.
Note also that if the court incorrectly estimates the value of performance to the
breach-against party, then the conclusion regarding the effects of the expectation
remedy on the breach decision and on the reliance decision would have to be
modified. Similarly, if the breached-against
party’s reliance expenditure are likely to be incorrectly determined, the conclusions
regarding the effects of the reliance remedy also would be different.
The discussion in this chapter has shown that, in general, there does not exist a
breach of contract remedy that is efficient with respect to both the breach decision
and the reliance decision. With respect to breach, the expectation remedy is ideal,
whereas with respect to reliance, the restitution remedy is ideal. Thus, which remedy
is best overall depends on whether the breach decision or the reliance decision is
more important in terms f efficiency. For instance, in the example used in this chapter
an inefficient breach occurred when S sold the widget to B2 when B2’s value was
\$180. Because, B1valued the widget at \$200, there was an efficiency loss of \$20
from the inefficient breach; assuming this occurs with a one-third chance (the chance
that B2 values the widget at \$180), the expected value of the efficiency loss
is\$6.67(1/3x\$20).
Inefficient reliance occurred when B1 spent the additional \$24 in reliance. Because
the expected benefit of reliance only was \$20, there was an efficiency loss from
inefficient reliance, the expectation remedy would be preferred.
A key assumption in the discussion in this chart was that the parties were
neutral with regard to risk. In chapter 8 we will reconsider breach of contract
remedies when the parties are assumed to be averse to risk and see that, in general,
none of the remedies discussed here are ideal with respect to risk allocation.

31

*21. The supposition that S breaches only when it is efficient for a breach to
occur is clearly counterfactual in general under the restitution remedy. This
supposition is used nonetheless because it allows the effect of the restitution remedy
on the extent of the reliance investment to be illustrated most easily.
*22. The “remedy” of no damages at all also would lead to the efficient
reliance investment for the same reason.
43
CHAPTER 6
THIRD APPLICATION—
AUTOMOBILE ACCIDENT
In both the applications discussed thus far—nuisance law and breach of
contract—it was appropriate to consider the possibility that bargaining among
the parties could lead to the efficient solution. Thus, the framework of the
Coase Theorem was directly applicable to these kinds of disputes. In the next
application that we will examine—automobile accidents involving pedestrians
—bargaining obviously cannot lead to the efficient outcome because neither
drivers nor pedestrians know in advance with whom to bargain. The Coase
theorem may be helpful nonetheless. Efficient legal rules for dealing with
driver-pedestrian accidents still can be derived by imagining what rules a driver
and a pedestrian would have chosen if they could have costlessly gotten
together before the accident. As in the other applications, the parties would
have agreed to remedies that lead them to behave so as to maximize their joint
benefits net of their joint costs.
A simple example will be used to investigate the efficiency of different
legal remedies in driver-pedestrian accidents. In this example, it is assumed
that the drivers and pedestrians are risk neutral; the discussion therefore will be
in terms of the expected accident cost to a pedestrian—the magnitude of the
harm if an accident occurs multiplied by its probability of occurrence. It also
is assumed initially that only the speed of drivers affects the pedestrians’
expected harm. (The example will be extended later in this chapter to include
the possibilities that the number of miles driven or the care exercised by
pedestrians also affect the expected harm.)
TABLE 3
Automobile Accident Example – Driver’s Care Affects Expected
Accident Cost
__________________________________________________________
__________

32

The driver has three choices: drive rapidly, drive moderately, or drive slowly.
Each choice results in some benefits to the driver and some expected accident
cost to the pedestrian.
The driver’s benefit from driving faster might be the dollar value he places on
saving time.
The pedestrian’s harm also is assumed to have a monetary value.*23
The data for the example are described in Table 3. For each choice of
the driver, the table lists the benefit to the driver and the expected accident cost
to the pedestrian. The efficient outcome requires that the driver act so as to
maximize benefit less cost. Given the data in table 3, it is efficient for the
driver to drive moderately. Relative to this outcome,
driving rapidly is inefficient because it increases the pedestrian’s expected
losses by \$60 while increasing the driver’s benefits only by \$40. And driving
slowly is inefficient because it lowers the driver’s benefits by \$30 while
lowering the pedestrian’s expected losses only by \$20.
*23. As suggested in note 6 above and in the accompanying text, economic
analysis still can be used when the harm is not equivalent to the loss of money
(as is the case with pain and suffering). However the discussion would be
more complicated.
45
The Driver’s Care
We will now consider the effects on the driver’s behavior of two
alternative rules of liability in accident law—strict liability and negligence.
Under each the driver will choose the action that maximizes his benefits net of
his expected liability payments. Under the rule of strict liability, the driver will
be made liable for the pedestrian’s accident losses regardless
of the driver’s care. Thus, for each action, the driver’s benefit net of his
expected liability payments is the same as the last column in Table 3. The
driver therefore will choose to drive moderately—the efficient outcome. In
essence, the rule of strict liability results in efficient behavior because it forces
the injurer—in this example, the driver—to take into account all of the adverse
effects of his behavior on the victim—the pedestrian.

33

For the rule of strict liability to be efficient, the court generally must be
able to obtain correct information about the victim’s damages. To see why, suppose
in the example that the court estimates damages to be one-half of the victim’s actual
damages. Then referring to table 3, the driver’s benefits net of expected liability
payments would be \$70 if he drives rapidly (\$120-\$50), \$60 if he drives moderately
(\$80-\$20), and \$40 if he drives slowly (\$50-\$10). He therefore would choose to
drive rapidly—faster than is efficient.*24 Conversely, suppose the court estimates
damages to be twice what they actually are. Then the driver’s benefits net of his
expected liability payments would be, respectively, --\$80 (\$120-\$200), \$0 (\$80-\$80),
and \$10 (\$50-\$40). Thus, the driver would choose to drive slowly—too slow relative
to desired driving behavior. In order to focus on other considerations, it will be
assumed hereafter that the court has accurate information about the victim’s damages.
Under the rules of negligence, the driver will made liable for the pedestrian’s
accident losses only if the driver does not meet some standard of care. Suppose this
standard is determined by the care that would be taken if the driver acted efficiently.
In the example, this corresponds to driving at moderate speed. Thus, the driver
would be liable for the pedestrian’s accident losses only if the driver chooses to drive
rapidly. Therefore, if he drives rapidly his benefit net of his expected liability
payments is \$20 (a\$120 benefit less a \$100 expected liability payment). If he drives
moderately, it is \$80 (just the benefit because there is no liability), and if he drives
slowly it is \$50
(again, just the benefit). Consequently, under the rule of negligence with this standard
of care, the driver will choose the efficient outcome of driving moderately. In
essence, the rule of negligence leads to the efficient outcome because the injurer
prefers to comply with the standard of care—to avoid having liability increase from
zero to the victim’s damages if the standard is violated—and the standard is selected
to correspond to the desired behavior.
For the rule of negligence to be efficient, it is necessary for the court to have
enough information to determine the efficient outcome so that the standard of care
can be chosen to correspond to it. To see why, suppose in the example that the court
mistakenly believes that it is efficient for the driver to drive slowly and therefore
makes this behavior the standard of care In other words the driver is liable for the
pedestrian’s losses if he drives rapidly or moderately, but not if he drives slowly.
Then referring to Table 3, the driver’s benefit net of his expected liability payments is
\$20 if he drives rapidly (\$120-\$100), \$0 if he drives moderately (\$80-\$40), and \$50 if
he drives slowly (\$50-\$0). Thus, the driver would choose to drive, an inefficient
outcome. Similarly, if the court were to make the standard of care too lenient rather
than too strict, the driver would choose to drive faster than would be efficient. It will
be assumed hereafter that the court has enough information to select the standard of
care that corresponds to the efficient outcome.
*24. For analogous reasons, the driver also generally would drive faster than is
efficient if, given his income or wealth, he does not expect to be able to pay the full
amount of the victim’s damages.

34

46/47
The discussion thus far illustrates a general principle in the economic analysis
of accident law: In accident situations in which the only issue is how to induce the
injurer to take appropriate care, both strict liability and negligence are efficient,
provided that liability equals actual damages if strict liability is used and that the
standard of care corresponds to the efficient outcome if negligence is used.
The Pedestrian’s Care
In many accident situations, however the problem is not just to control the
injurer’s behavior. In general, both the injurer and the victim can affect the
probability and the magnitude of the harm. For example, a pedestrian can walk rather
than run while crossing a street, or a cyclist can wear a protective helmet. When both
the injurer and the victim can affect the expected hard, the issue is how to induce both
parties to take appropriate care. We will now reexamine the rules of strict liability and
negligence with respect to this additional consideration.
To allow for the expected harm to be determined by the behavior of both the
driver and the pedestrian, it is necessary to extend the example used above. It will
now be assumed that the pedestrian has one choice—whether to walk or to run. If she
walks, then her expected accident loss is \$100 if the driver drives rapidly, \$40 if the
driver drives moderately, and \$20 if the driver drives slowly. These are the same
values used in Table 3. If the pedestrian runs, her corresponding expected accident
lasses are \$110, \$50, and \$30. In other words, running is assumed to raise the
expected harm by \$10 regardless of the driver’s behavior.*25 The data for the
extended example are summarized in table 4, where it also is assumed that the
driver’s benefits from driving are the same as in Table 3.
The efficient solution to the accident problem now involves a specific action
both by the driver and the pedestrian. If the pedestrian walks, the problem is the same
as before, and the efficient outcome with respect to the driver’s behavior is from him
to driver moderately. If the pedestrian runs, Table 4 shows that benefits minus costs
also are maximized when the driver drives

35

*25. In general, the effect of the pedestrian’s care on expected accident loses
would depend on the driver’s behavior. For example, suppose the pedestrian’s
decision
whether to walk or to run determines the probability of an accident, while the
driver’s speed
determines the magnitude of the harm if an accident occurs. Then the faster the
driver drives, the more the expected harm will be raised by the pedestrian’s
decision to run. The assumption made in the text—that running raises the
expected harm by an amount that does not depend n the driver’s behavior—
simplifies the subsequent analysis without affecting the general conclusions.
48/49
Table 4
Automobile Accident Example—Driver’s Care and Pedestrian’s
Care
Affect Expected Accident Cost

Behavior
of
Drivers

Benefit
To
Driver

Drive rapidly

\$120

Drive moderately

\$80

Drives Slowly

\$50

Expected
Accident Cost
to Pedestrian
(Depending on
Pedestrian’s
Behavior)
\$100(walks)
\$110(runs)
\$40(walks)
\$50(runs)
\$20(walks)
\$30(runs)

Benefit Minus
Cost
(Depending
on
Pedestrian’s
Behavior)
\$20(walks)
\$10(runs)
\$40(walks)
\$30(runs)
\$30(walks)
\$20(runs)

moderately. Thus, regardless of the pedestrian’s behavior, the efficient solution
involves driving moderately. Whether it is efficient for the pedestrian to walk or to
run depends on the relevant costs and benefits. Running rather than walking increases
the pedestrian’s expected harm by \$10 (regardless of the driver’s behavior). It will be
assumed that running provides additional; benefits to the pedestrian valued at \$5—for
example, due to the saving of time. Thus, given these costs and benefits, the efficient
solution involves the pedestrian walking. *26
Now reconsider the rule of strict liability. The driver’s benefit net of his
expected liability payments corresponds to the last column in Table 4. If the
pedestrian walks, the relevant values are \$20, \$40, and \$30, depending on whether
the driver drives rapidly, moderately, or slowly. The driver therefore would chose to
drive moderately. If the pedestrian runs, the corresponding values are \$10, \$30, and
\$20, and the driver also would choose to drive moderately. Thus regardless of the
pedestrian’s behavior, the rule of strict liability will lead the driver to behave
efficiently in this example. However, the rule of strict liability will not be efficient
with respect to the pedestrian’s behavior. Because the pedestrian will be fully

36

compensated for her losses, she will ignore these losses when deciding whether to
walk or to run. She only will consider the \$5 extra benefit from running. The
pedestrian therefore will choose to run even though running increases expected
accident costs by \$10.
The problem of controlling the victim’s behavior under the rule of strict
liability can be solved by adding a defense of contributory negligence. Then, the
injurer is strictly liable unless the victim is contributory negligent. This rule will
result in the desired behavior of both parties.
To see this in the example, let the standard of care applicable to the pedestrian
correspond to the efficient behavior of the pedestrian—walking. Thus, if the
pedestrian walks, she is not contributory negligent, so the driver would be strictly
liable. Is she runs, she is contributory negligent so the driver would be free of
liability. The pedestrian then has to bear her own losses.
Thus, while running rather than walking provides benefits valued at \$5, it increases
the expected
Accident cost borne by the pedestrian from zero to \$110, \$50, or \$30, depending on
whether the driver drives rapidly, moderately, or slowly (see Table 4). Clearly, the
pedestrian will choose to walk in order to avoid having to bear her own losses. Given
this choice by the pedestrian, the driver will be strictly liable. We already have seen
that this will lead the driver to choose to drive moderately. Thus, the rule of strict
liability with a defense of contributory negligence will cause both parties to take an
efficient amount of care.
Next, reconsider the rule of negligence in terms of the incentives it creates for
both parties to take appropriate care. Assume, as before, that the driver is negligent
only if he drives rapidly. If the pedestrian walks, the driver’s benefits net of his
expected liability payments are the same as discussed earlier under the negligence
rule, so the driver will choose to drive moderately.
*26. For simplicity, the benefit to the pedestrian from running or walking is
not include in Table 4. Thus, Table 4 focuses on the benefits and costs
associated with the driver’s behavior (which depend in part on the pedestrian’s
behavior).
50
If the pedestrian runs, the driver’s benefits net of his expected liability
payments ar \$10 if he drives rapidly (\$120--\$110), \$80 if he drives moderately
9\$80--\$0), and \$50 if he drives slowly (\$50--\$0). Thus, the driver will choose
to drive moderately regardless of what the pedestrian does. Because the driver
therefore will not be negligent, the pedestrian will bear her own losses. She
will then compare the \$5 extra benefits from running to the \$10 increase in
expected accident costs and will choose to walk. Thus, the rule of negligence
will lead both parties to take an efficient amount of care.
Note that under the negligence rule it is not necessary to add a defense of
contributory negligence to induce the victim to take proper precautions. If a

37

contributory
negligence defense were added, it would not be affect the conclusion that both
parties will take an efficient amount of care. The victim would meet the
standard of care applied to her to avoid being contributorily negligent and
having to bear her own losses. Given that the victim is not contributorily
negligent, the injurer will meet the standard of care applied t him to avoid
being negligent and having to compensate the victim for her losses.
The preceding discussion of the accident problem when both parties can
affect the expected harm illustrates another general result in the economic
analysis of liability rules: in accident situations in which the problem is to
induce both the injurer and the victim to take appropriate care, a rule of strict
liability with defense of contributory negligence or a rule of negligence—with
or without a defense of contributory negligence—is efficient.
The Activity-Level Issue
In many accident situations, expected accident losses depends not only
on the care exercised by each party, but also on the extent to which each party
participates in the activity that is the source of the dispute. For example the
number of driver-pedestrian and on how frequently pedestrians travel by foot
(rather than, say, by bus). The efficient level of participation in the disputecreating activity is determined by comparing the benefits a party obtains from
greater participation—for example, from the more extensive use of one’s car—
to the resulting increase in expected accident costs. In general, then the
problem to be solved by liability rules is how to induce both parties to take
appropriate care and to engage in the activity to an appropriate extent.
To examine whether the rules of strict liability and negligence lead to the
efficient levelof participation in the activity, the simple version of the driverpedestrian example—in which only the driver’s speed affects the pedestrian’s
expected accident costs—will be extended to include the number of miles
driven. (The more general case in which expected losses also are affected by
the pedestrian’s care and level of participation in the activity will be discussed
below.) Now suppose that expected accident costs depend not only on whether
the driver drives slowly, moderately, or rapidly, but also on whether he drives
“a little”
or ”a lot”. If he drives a little, then the relevant data are assumed to be the
same as in the simple version of the example—that is, the same as in Table 3.
If he drives a lot, then his benefits are assumed to increase by \$20 and the
pedestrian’s expected accident costs are assumed to rise by \$30, regardless of
the speed at which he drives.*27 The data for the extended example are
summarized in Table 5.
Because the additional benefits from driving a lot are less than the increase in
expected accident costs, the efficient solution involves driving a little. If the
driver drives only a little, the problem is the same as that discussed in the

38

simple version of the driver-pedestrian example, where it was efficient for the
driver to driver to drive at moderate speed. Put differently, driving moderately
and only a little maximizes benefits less costs. This can be seen directly in the
last column of Table 5.
*27. A point analogous to the one made in note 25 above applies here
TABLE 5
Automobile Accident Example—Driver’s Care and Activity Level
Affect Expect Accident Cost
_______________________________________________________________
________

52

Thus, the driver will choose to drive moderately and only la little. As in the
earlier versions of the driver-pedestrian example, strict liability leads the
injurer to behave efficiently because it forces him to take into account the
adverse effects of his behavior on the victim. The only difference now is that
one relevant aspect of his behavior is the extent of his participation in the
activity.
Under negligence, suppose, as in the earlier versions of the example, that
the driver is negligent only if he drives rapidly. Then, if his participation in the
activity corresponds to driving a little, his benefit net of his expected liability
payment is \$20 (\$120-\$100) if he drives rapidly, \$80 (\$80-\$0) if he drives
moderately, and \$50 (\$50-\$0) if he drives slowly.
If he drives a lot, the comparable values are \$10 (\$140-\$130) if he drives
rapidly, \$100 (\$100-\$0) if he drives moderately, and \$70 (\$70-\$0) if he drives
slowly. The driver therefore will choose to drive a lot and to drive at moderate
speed. In other words, the negligence rule with this standard of care is efficient

39

with respect to the injurer’s care but not
with respect to his level of participation in the activity.
Recall from the discussion of the negligence rule in the simple version of
the driver-pedestrian example that this rule is efficient behavior of the
injurer.*28 The negligence rule is not efficient in the present version of the
example precisely because the standard of care does not take into account one
relevant aspect of the injurer’s behavior—the extent of his participation in the
activity. If the standard of care were to correspond to the efficient outcome of
driving moderately ad only a little.—so that the driver would be negligent if he
drives a lot even if he drives moderately—then the negligence rule also would
*28. See pp.45/46 above
In practice, however, it often is not feasible to include the level of
participation in the activity as an aspect of the standard of care. For example, it
would be virtually impossible for a court to determine how many miles a
particular person drives each year since that person might drive a car that is
shared with other family members or he might drive different cars owned by
household. If the injurer’s level of participation in the activity is omitted from
the standard of care, than a negligence rule generally will lead him to
participate in the activity to an excessive degree. The reason for this is
straightforward. If the care he exercises meets the standard of care, he will not
be liable for any damages. Therefore, in deciding how much to participate in
the activity, he will consider the additional benefits from greater participation
but not the increase in expected accident costs. This problem with the
negligence rule was illustrated by the driver-pedestrian example. Given a
standard of care based only on the driver’s speed, the driver chose to meet the
standard by driving moderately. But he also chose to drive a lot, exceeding the
efficient level of participation in the activity.
The discussion thus far of the activity-level issue can be summarized as
follows: In accident situations in which the problem is to induce the injurer
both to take appropriate care and to participate in the activity at an appropriate
level, strict liability is efficient. Negligence also is efficient if the standard of
care encompasses both the injurer’s care and his level of participation in the
activity. However, if the standard does not include the injurer’s activity level,
then the negligence rule will lead to excessive participation in the activity. In
practice, the negligence rule is likely to be inefficient for this reason.
In some accident situations, the expected accident losses may depend not
only on the injurer’s care and activity level, but also on the victim’s care and
activity level. In this more general setting, results analogous to those just
discussed would occur. Strict liability with a defense of contributory
negligence would be efficient if the standard of care applicable to the victim

40

encompasses both the victim’s care and her activity level. However, if it
includes only her care, then the victim will engage in the activity to an
excessive degree. Similarly, negligence would be efficient if the standard of
care applicable to the injurer includes both aspects of his behavior. If it
includes only his care, then he will participates in the activity too much. Thus,
if it is not feasible to include either party’s activity level in the standard of care,
the preferred liability rule depends on whether it is more important to control
the injurer’s or the victim’s activity level. If the injurer’s activity level is of
greater concern, then strict liability with a defense of contributory negligence
should be used. If the victim’s activity level is more important, then
negligence is preferable.
The final consideration in the economic analysis of liability rules that
will be discussed in this chapter is the effect of each rules that will be discussed
in this chapter is the effect of each rule on the administrative costs of resolving
accident disputes. These costs depends both on the number of cases litigated
and on the costs of resolving each case.*29 The negligence rule would be
expected to generate less litigation than the strict liability rule for the following
reason.

*28. See pp. 45-46 above
*29. Although the following discussion is concerned with disputes that are
litigated, similar points could be made with respect to disputes that are settled
after costly negotiation.

55
Consider the negligence rule is an accident situation in which the injurer’s care
was very likely to have satisfied the standard of care. Given the cost to the
victim of litigating, she might not find it worthwhile to bring an action against
the injurer because of the low probability of success. Yet, under the strict
liability rule, she might be willing to bring an action in this accident situation
because the injurer’s care is not a bar to a successful suit.
Although there may be fewer cases under the negligence rule, the
administrative cost of resolving each one is likely to be higher than under the
strict liability rule. The justification for this conclusion is easiest to see when
the only problem is to induce the injurer to take appropriate care. To apply a

41

strict liability rule, the court needs to know the victim’s damages. But to apply
a negligence rule, the court not only needs to know the victim’s damages and
the injurer’s benefits at different levels of the injurer’s care—in order to choose
a standard of care that corresponds to the efficient outcome—but also how the
injurer behaved—in order to determine whether he met the standard. (When the
problem is to introduce both the injurer and the victim to take appropriate care,
this argument does not apply because the strict liability rule then requires a
defense of contributory negligence to be efficient.)
On balance, therefore, administrative cost considerations do not clearly
favor either rule. While the negligence rule is likely to lead to fewer cases
being litigated than the strict liability rule, it may well generate higher
This chapter has shown that in accident situations in which the only
problem is to create incentives for the parties to take appropriate care, both
strict liability with a defense of contributory negligence and negligence are
efficient. If the problem also is to induce the parties to engage in the disputecreating activity to an appropriate extent, then both rules still are efficient
provided that the relevant standards of care—the victim’s under the
contributory negligence defense and the injurer’s under the negligence rule—
incorporate the activity-level decision of the party to whom the standard
applies. In practice, however, this is not likely to be feasible. If the standard of
care refers only to the relevant party’s level of care, then strict liability with a
defense of contributory negligence will lead to excessive participation in the
activity by the victim, and negligence will lead to excessive participation by
the injurer. In many accidents situations it may be apparent that one party’s
activity level matters more than the others, in which case the superiority of one
of the rules will be clear.
An important assumption in the discussion in this chapter was that both
drivers and pedestrians were risk neutral. When accident law is reconsidered in
Chapter 9 under the assumption of risk aversion, it will be seen that strict
liability and negligence may no longer be efficient even in accident situations
in which only the injurer’s care determines the victim’s expected losses.

42

57
CHAPTER 7
RISK BEARING AND INSURANCE
In both the breach of contract and automobile accident applications,
some sort of risk was present. In the contract example, there was uncertainty
about the needs of the second buyer; and in the accident example, there was,
implicitly, uncertainty regarding whether n accident would occur. Because of
the assumption in both cases that the parties were risk neutral, risk per se did
not matter.
We will now consider the generally more realistic assumption that parties
are
risk averse (at least with respect to large risks). This means that they care not
only about the expected value of a risky situation, but also about the absolute
magnitude of the risk. For example, a risk-averse person, unlike a risk-neutral
person, would not be indifferent between the certainty of winning \$5000 and a
\$50 percent chance of winning \$10,000. The risk-averse person would, by
definition, prefer \$5,000 with certainty.*30
The Bearing of Risk
To see the value of reducing or eliminating the risk borne by risk-adverse
persons, consider the following situation. Suppose you have recently
graduated from law school and have become an associate in a private law firm.
Your first assignment is to work full-time on a case that the firm has accepted
on a contingent fee basis—the firm will receive \$200,000 at the end of the year
if it wins the case and nothing otherwise. Your employer knows from previous
experiences that there is a 50 percent chance to winning the case.
The firm proposes to pay you \$200,000 for the year if you win and nothing if
you lose, so your expected salary is \$100,000. How much would you be
willing to accept with certainty instead? Suppose you would accept little as
\$80,000 to avoid being subjected to the risk.
This situation provides a nice business opportunity for me. I am going to
start a business with some friends in which we will bear the risk of income
fluctuations of recent law school graduates who work on contingent fee cases.
Here’s our offer: We will pay you \$90,000 with certainty if you assign to us the
claim on the law firm you are working for. Suppose his offer attracts one
hundred other lawyers in circumstances similar to your. At the end of the year
we can count on approximately fifty of our claims to pay \$200,000 and the rest
nothing, so our expected revenue is \$10 million. Because we have to pay one
hundred lawyers \$90,000 each, our costs are \$9 million. Thus our profit will be
approximately \$1 million. We are better off entering into this deal—although
each of us is risky averse too, our risks are minimal since we can be confident

43

that approximately fifty of our claims will pay off.*31 You are better off
under this deal too—you were willing to accept as little as \$80,000 to avoid the
risk, but you have received \$90,000. Therefore, this agreement between us—
which is a form of insurance—is efficient.
If it isn’t already obvious, we should now identify ourselves. We are the
partners in your law firm. The partners bear the risks of the firm‘s successes or
failures, while the associates’ incomes are guaranteed. This make sense
because the partners can better bear the risks of the firm—they can ”average
out” the results of many risky cases and , because they have more wealth, they
can better absorb the risks that remain.*32

*30. Individuals may also be risk preferring. In the example in the text this
would mean that they would prefer a 50% chance of winning \$10,000 to the
certainty of winning \$5,000.
This case will not be dealt with in the book.
The preceding discussion has been about a beneficial risk—that is a risk
that someone would accept voluntarily. For such risks, a risk-averse person is
willing to settle for less than the expected value of the risk. In the law firm
example, rather than accept the
Risk of an equal chance of receiving \$200,000 or nothing, with an expected
benefit of \$100,000, the associate was willing to accept as little as \$80,000
with certainty. Similarly, for a detrimental risk—a risk that someone would not
accept voluntarily—a risk-averse person would be willing to pay more than the
expected value of the risk to avoid it. For example, consider a risk involving
an equal chance of losing \$200,000 or nothing, thus, having an expected loss of
\$100,000. A risk-averse person might be willing to pay as much as \$130,000
to avoid this risk.
Because risk-averse individuals are willing to pay for the reduction of risk, just
as they are willing to pay for more tangible commodities, the benefit from
eliminating or reducing the risk borne by such individuals is properly included
in the efficiency calculus. For a beneficial risk, a natural way to measure the
benefit from eliminating risk is to compare the expected value of the risk and
the least amount of money the person would accept with certainty instead. In
the law firm example, the expected value of the risk was \$100, 000 while the
associate was willing to accept as little as \$80,000 with certainty. Thus, the
benefit from removing the risk was \$20,000. Similarly, for a detrimental risk,
the benefit from eliminating risk can be measured by the difference between
the expected value of the risk and the greatest amount of money the person
would pay to avoid the risk. Thus, in the detrimental risk example at the end of
the previous paragraph, the benefit from removing the risk would be \$30,000
because the risk-averse person was willing to pay \$130,000 to avoid a risk that
had an expected loss of \$110,000.

44

*31. Readers familiar with the basic principles of probability theory will
recognize that the conclusion that our risks are minimal does not follow
without some additional assumptions. It is sufficient to assume that the
outcomes of the associates’ cases are “independent” (in the technical statistical
sense) and that each of our shares in the business is “small”.
*32. This statement obviously presumes that the higher a person’s wealth, the
less averse she is to a given size risk. This is a standard assumption in the
economic analysis of risk.

60
Insurance
One commune way of eliminating risk is through insurance. For
detrimental risks, the insured person pays some amount of money with
certainty—the insurance premium—in return for which she is full compensated
if the undesirable risk materializes. For beneficial risks, the insured person
receives some amount of money with certainty in return for allowing someone
else to benefit from the desirable outcome if it occurs, as in the law firm
example. This too may be thought of as a form of insurance.
An insurance policy that completely eliminates risk might, however,
have an undesirable side effect. In the law firm example, the associate will
have less of an incentive to work hard in order to win the case if her income for

45

the year does not depend on her winning the case. This kind of problem also
arises with respect to undesirable risks. For instance, if personal items left in
your car are completely insured against theft, you might be more likely to leave
your camera on the back seat rather than to go to the trouble of putting it in the
trunk of the car. This example illustrates a general problem: the provision of
insurance tends to increase the probability of a loss or the size of the loss
because the insured person has less of an incentive to take precautions. This
phenomenon is referred to in the insurance literature as a problem of moral
hazard.*33
In principle, the moral hazard problem can be overcome by adjusting the
insurance premium to reflect the increase in the expected loss returning form
the insured person’s taking less care. For example, suppose your camera is
worth \$500 and that putting it in the trunk of your car eliminates the possibility
of theft, whereas leaving it on the back seat leads to a one-in-a-hundred chance
of having it stolen. In other words, leaving it on the back seat results in an
expected loss of \$5. If your insurance premium were to increase by \$5 if you
regularly left your camera on the back seat of your car, then you would leave it
there only if it is worth at least \$5 to you to do so. It may or may not be worth
it for you to leave it there. For example,, if you are a professional
photographer specializing in outdoor photography, it probably would be worth
paying \$5 more for the extra convenience, whereas for most people the added
convenience would not be worth this much. By being forced to pay more
because of the increased expected loss, the insured person will have the
appropriate incentive to take precautions. In other words, the moral hazard
problem can be eliminated if the insurance premium depends on how much
care is exercised by the insured person.
This solution to the moral hazard problem often is not feasible in
practice because the insurer cannot cheaply monitor the behavior of the person
being insured. In the camera example, to decide whether to raise the premium
by \$5, the insurance company would have to determine whether the insured
person regularly left her camera on the back seat of her car. In the law firm
example, monitoring of associates’ effort’s easier, and as a result, bonuses and
promotions can be used to some extent to encourage appropriate effort.
There are other alternatives to monitoring the insured person’s behavior
only partial insurance coverage in order to induce the insured person to take
some precautions. Sometimes this partial coverage takes the form of a
deductible, in which, for example, the insured person bears the first \$100 of
loss and the insurance company bears the rest. Other times partial coverage
takes the form of co-insurance, in which, for intense, the insured person bears
20% of all losses. In either case this approach is obviously a compromise
because it leaves some risk on a risk-averse person and it will not completely
solve the problem of moral hazard. On balance, however, partial insurance
maybe preferable both to no insurance at all (which leaves the most risk on

46

risk-averse persons) and to complete insurance (which provides little or no
incentive to take precautions).
62
The discussion in this chapter has shown that the elimination of
reduction of risk borne by a risk-averse person is a benefit that should be taken
into account in applying the efficiency criterion. One common way in which
risk is reduced is through insurance. An ideal insurance policy has two features
—it provides full coverage in order to eliminate the bearing of risk, and it bases
the premium on the behavior of the insured person in order to avoid the
problem of moral hazard. In practice, however, monitoring the insured
person’s behavior often is difficult , in which case it may not be possible to
base the premium on the insured’s behavior. Then it may be desirable to
provide only partial insurance coverage in order to increase insured person’s
incentive to take precautions.
We will now return to the breach of contract and automobile application
and reexamined them in the light of these principles of risk bearing and
insurance. The contract application illustrates he case of the beneficial risk
(uncertainty about a higher third-party offer), while the accident application
obviously represents the case of a detrimental risk.

*33.

The previous sentence illustrates the problem of moral hazard with

47

respect to a detrimental risk, but moral hazard also can occur when the risk is a
beneficial one, as in the law firm example. Then the provision of the insurance
tends to decrease the likelihood of a gain or the size of the gain because the
insured person has less of an incentive to engage in the effort to obtain the
gain.
63
CHAPTER 8
FORTH APPLICATION –
BREACH OF CONTRACT AGAIN
In the discussion of breaches of contrasts in Chapter 5, it was assumed that the
parties were risk neutral. The primary conclusions there were that the
expectation remedy was most efficient with respect to the breach decision and
that the restitution remedy was most efficient with respect to the reliance
decision. We will now examine breach of contract remedies when at least one
of the parties is risk averse. In order to focus on the risk-allocation issue,
assumption will be made that imply that both the breach decision and the
reliance decision will be efficient under all of the remedies considered.
The analysis will be undertaken using a variation of the earlier example
of a seller S, who can produce a widget for \$150, an initial buyer B1, who
values the widget at \$200 and who had to make a reliance investment of \$10,
and a possible second buyer B2. The contract price agreed to by S and B1is
again payable in advance.*34 It is assumed in this chapter that B1’s reliance
expenditure is fixed, so there cannot be a problem of inefficient reliance. It is
also assumed that the value B2 attaches to the widget is either \$0 or \$250. The
fact of this example are summarized in Table 6 (which is a slightly modified
version of Table 2).
34. Although none is the results in this chapter depend on the specific contract price, it will
be useful to keep in mind that the contract price varies with the damage payment for the reasons
discussed in Chapter 5. See pp. 32-33 above.

48

TABLE 6
Breach of Contract Example
S is the seller:
S’s cost to produce the widget is \$150.
B1 values the widget at \$200.
B1’s reliance expenditure is \$10.
B1 pays S the contract price in advance.
B2 values the widget at \$0 or \$250.
By eliminating the possibility considered earlier that B2’s value might be \$180,
the problem of inefficient breach is avoided under all of the remedies examined; this
can be explained as follows. It was shown in Chapter 5 that the level of damages
under the remedies considered was at least as large as the contract price (this was the
level of damages under the restitution remedy), but no greater than the \$200 value B1
attaches to the widget ( the level of damages under the expectation remedy). 35
Consequently, if B2 does not need the widget, S obviously will not breach, while if
B2’s offer is \$250, S will breach in order to sell to B2 regardless of which remedy is
used. In other words, the breach decision will be efficient under all of the remedies
considered. Thus, since the reliance expenditure is fixed and the breach decision is
efficient, the only issue to be discussed is risk allocation.
35. See pp. 34-35 above

Optimal Risk Allocation
Because it will be assumed throughout this chapter that private insurance is not
available to the parties, the allocation of the contract risks will be determined solely
by the remedy for breach of contract. Before examining the effects of different
remedies, it will be used to determine the level of damages that would be awarded in
the widget example if the risk were allocated efficiently or optimally. Suppose for a
start that the buyer B1 is risk averse and that the seller S is risk neutral.*36 Then,
because it is desirable to eliminate the risk imposed on risk-averse persons, B1
should, in effect, be insured and S should bear all of the risk. This can be

49

accomplished by making S pay B1 in the event of a breach an amount of money equal
to the value B1 attaches to the widget – that is, \$200. Given this damage payment,
B1’s profit equals her \$200 benefit less the contract price and less her \$10 reliance
expenditure regardless of whether B2’s offer materializes and S breaches. S’s profit,
however, does depend on whether B2’s offer occurs. If the offer does not occur,
S’s profit equals the contract price less his \$150 production cost, whereas if the
offer does occur, his profit is augmented by \$50 – the difference between B2’s \$250
offer and the \$200 damage payment to B1. Thus, with this damage payment, B1 does
not bear any risk but S does.
Now suppose that S is risk averse and B1 is risk neutral. Then S should be
insured and all of the risk should be borne by B1. This can be achieved by making S
pay B1, if a breach occurs, an amount of money equal to B2’s offer—that is \$250.
S’s profit then equals the contract price less his \$150 production cost regardless of
whether the contract is performed or breached because, in the event of a breach, he
must disgorge the extra profits he otherwise would have obtained from the sale to B2.
B1’s profit equals her \$200 value less the contract price and less her \$10 reliance
contract is performed or the \$250 damage payment less the contract price and
reliance expenditure if the contract is breached. Thus, only B1’s profit is uncertain.
Finally, if both B1 and S are risk averse, the risk should be shared between
them in a way that reflects their relative aversion to risk. This can be accomplished
by a damage payment between the two extremes discussed above—that is , between
B1’s value of \$200 and B2’s offer of \$250, B1 bears the risk of a higher offer because
the difference between B1’s profit when the contract is performed and B1’s profit
when the contract is breached increases. Similarly, to the extent that the damage
payment is below\$250, S bears this risk because S’s profit becomes more uncertain.
Thus, the more risk averse B1 is relative to S, the lower the optimal damage payment.
Note, however, that the optimal damage payment is never below the \$200 value B1
attaches to the widget. It equals this value only when B1 is risk averse and S is risk
neutral.

*36. Recall from note 19 that B2 is assumed to offer an amount equal to the value he
attaches to the widget, so his profits do not depend on how valuable the widget is to
him. In other words, B2does not bear any risk. This is why it does not matter in the

50

example whether B2 is risk neutral or risk averse.
THE EFFECT OF THE REMEDIES

We can now reexamine the expectation, reliance, and restitution remedies to
see whether, and under what circumstances, they optimally allocate the contract risks
in the example. Under the expectation remedy, if the seller breaches, the buyer can
recover from the seller an amount of money that puts the buyer in the same position
she would have been in had the contract been completed. As seen in Chapter 5, this
corresponds in the example to a damage payment equal to the \$200 value B1attaches
to the widget.*37 B1’s Profit then does not depend on whether S breaches to sell the
widget to B2, but S’s profit does. In other words the beneficial risk of B2’s offer is
borne entirely by S; B1 is in effect completely insured against the risk…Based on the
early discussion of optimal risk allocation, we can conclude, therefore that the
expectation remedy is efficient in terms of risk allocation only if the buyer is risk
averse and the seller is risk neutral.
Under the reliance remedy the buyer can recover an amount of money that puts
her in the same position she would have been in had she never enter into the contract
with the seller. It was seen in Chapter 5 that reliance damages correspond in the
example to B1’s \$10 reliance expenditure plus the contract price; it also was
demonstrated that this level of damages was less than the level of expectation
damages.—that is, less than the \$200 value B1 attaches to the widget.*38 Recall
from the discussion of optimal risk allocation earlier in this chapter that the optimal
damage payment never is below the \$200 value B1attaches to the widget and that is
only this low when B1is risk averse and S is risk neutral. Thus, the reliance remedy
never leads to the optimal allocation of the contract risk in this example. Intuitively,
this is because the effect of the reliance remedy is to accentuate the risk created by
the possibility of a third-party offer. The further the damage payment falls below
B1’s value of the widget , the greater the variability of both B1’s profit and S’s profit.
Under the resolution remedy, the buyer can recover any benefit that she has
conferred on the seller. It was shown in Chapter 5 that restitution damages correspond
in the example to the contract price, which will be below B1’s \$200 value.*39
Thus the restitution remedy, like the reliance remedy, never would optimally allocate
the contract risk because the damage payment always would be too low.
The preceding discussion illustrates a general conclusion about the riskallocation effects of breach of contract remedies: When the contract risk is due to the
possibility of a third-party offer, the expectation remedy is ideal if the buyer is risk
averse and the seller is risk neutral. Moreover, it is preferable to both the reliance and
the restitution remedies (regardless of the parties’ relative aversion to risk).
in many contract situations, however, the seller may be risk averse and the buyer may
be risk neutral, in which case the parties would want the buyer to bear the risk; or
both parties might be risk averse, in which case they would want to share the risk
rather than to allocate it entirely to one party.
*37. See pp.33-34 above
*38. See pp.34-35 above

51

*39. See p. 35 above

Then the expectation remedy is not ideal. However, there is an alternative—a
liquidated damage remedy—that can allocate the contract risk between the parties in
any way they desire. Under a liquidate damage remedy if the seller breaches, the
buyer can recover an amount of money agreed to by the parties in advance. This
remedy differs from the others in that it is chosen by the parties rather than imposed
upon them by a court. If their concern is with the allocation of risk, as we are
assuming in this chapter, then they would choose a damage payment that allocates the
risk according to their relative aversion to risk. In principle, then a liquidated
damages